Evolution of cooperative behavior in the iterated prisoner's dilemma under random pairing in game playing

摘要

We discuss the evolution of cooperative behavior in the iterated prisoner's dilemma (IPD) game under random pairing in game playing. The main characteristic feature of this paper is the use of the random pairing scheme in which each player plays against a different randomly chosen opponent at every round of the dilemma game. Each player has a single-round memory strategy represented by a binary string of length five. The next action of a player is determined by its strategy based on the result of its previous round of the dilemma game. First we perform computational experiments to examine the evolution of cooperative behavior under the random pairing scheme using various parameter specifications. Experimental results show that the evolution of cooperative behavior is difficult independent of parameter specifications about the population size, the crossover probability, and the mutation probability. It is also shown that slightly better results (i.e., higher payoff) are obtained from smaller populations. Then we demonstrate the possibility of the evolution of cooperative behavior under the random matchmaking scheme in the case of a spatial IPD model where each player is located in a cell in a two-dimensional grid-world.